=> MOST RECENT POSTS ARE ON TOP
4. Setting C0=0
As suggested by Grell/Freitas/Nilo, I set c0=0 in file cup_up.f90 and commented out outtem() and outrt() in file cup_grell2.f90. These settings should forbid precipitation to occur and enforce mass conservation. If I understood correctly, the vertical integration of the tendency in q and qc should then add to zero (compensate each other). The histogram below shows this is not the case. The average negative value means the scheme is removing water from the column even thou there is not precipitation reaching the ground.
3. First diagnostics
I have modified file cup_grell2.f90 in order to do the vertical mass integration of the total moisture tendency. All modifications start with "!hmjb". Contributions from q and qc, and surface precipitation are printed separately into a file named 'fort.200'. I had to include USE statements for mem_scratch3_grell in subroutine CUPARTH to have access to variable p_cup.
For conserving the mass of water in the atmospheric column the mass of water removed (vertical integration of the tendency) must match that appearing as surface precipitation (variable 'pret'). The histogram below shows that this is not the case. It shows their difference to the average value. It means that, in a single time-step, in a single atmospheric column, the parameterization is unbalanced by -6% (removes more water than precipitates) to +4% (precipitates more than removed).
2. Testing Suite
For testing the water mass conservation BRAMS was integrated for 12 hours, starting at 12z 1st September 2002. Just one coarse grid (35 x 35) with 112km resolution. 32 vertical levels starting with 120m and 1.2 stretch ratio. All runs with 1 single processor (serial run). Important files are tabulated below:
| include.tgz | Original and modified intel include options. I had to add '-lsz' flag. Couldn't compile without that. |
| configure.tgz | Original and modified configure scripts. I had to add 'libsz.a'. Couldn't compile without that. |
| install.tgz | Original and modified install scripts. I had to add 'libsz.a' installation. Couldn't compile without that. |
| RAMSIN-grell | BRAMS Namelist |
| dprep_reference.tgz | dprep files used in the run |
| data_reference.tgz | Files generated inside data/ after the MAKESFC run |
| ivar_reference.tgz | Files generated inside ivar/ after the MAKEVFILE run |
1. Reference Code
The grell ensemble scheme as implemented in the official and currently stable release (4.2) is used as reference for these tests. The BRAMS-4.2/src/brams/cuparm directory is available: cuparm_reference.tgz.